When culture, serology, and polymerase chain reaction (PCR) can’t identify the cause of a patient’s infection, inappropriate therapy, excess healthcare costs, or even the individual’s death all are possibilities. In contrast to these tests that generally look for one pathogen at a time, metagenomic next-generation sequencing (mNGS) analyzes a broad spectrum of microorganisms at once, potentially providing quicker diagnoses and avoiding untoward outcomes.
A clinical mNGS test is expensive—sometimes more than $2,000. But it might be worthwhile when routine tests don’t provide information, for immunocompromised patients infected with pathogens that do not affect healthy people, or for patients who can’t tolerate invasive diagnostic procedures, said clinical laboratorians who use the test.
“It’s exciting to think about how mNGS can help us improve care. There’s a lot of promise in these methods,” said David Peaper, MD, PhD, an assistant professor of laboratory medicine at Yale University and director of clinical microbiology at Yale New Haven Hospital in New Haven, Connecticut. However, these tests are limited by risk of contamination, designs geared to specific types of pathogens, bioinformatics challenges, and databank deficiencies, said Peaper and others.
Diverse Pathogen Detectors
Clinical mNGS typically involves extracting cell-free (cf) DNA, cfRNA, or both from a body fluid, amplifying the nucleic acids via PCR, generating libraries, and shotgun sequencing nucleic acids at a very high depth. Genomic laboratories use software to analyze the millions of reads generated in each sample and identify those that align to nucleotide sequences of pathogens in various databases, such as the National Center for Biotechnology Information (NCBI) GenBank.
A $2,200 University of California, San Francisco (UCSF) test analyzes both DNA and RNA to diagnose causes of meningitis and encephalitis from bacteria, viruses, fungi, and parasites found in cerebrospinal fluid. UCSF software analyzes reads, identifies those that align to pathogens in GenBank, and issues a qualitative report noting the pathogens present in the sample, along with interpretive clinical notes. A sequencing board, modeled on a tumor board, discusses results in real time with treating physicians and may make recommendations about additional testing. Turnaround time from shipping samples to delivery of a qualitative report is generally within a week, said Charles Chiu, MD, PhD, a key developer of the UCSF test and a professor of laboratory medicine and director of the clinical microbiology laboratory at UCSF.
The test performs well in head-to-head comparisons with routine clinical testing and can make diagnoses usual microbiology tests cannot, according to recent research (NEJM 2019;380:2327–40). Among 204 pediatric and adult patients in eight hospitals, the UCSF test detected 58 infections in 57 of the patients. Hospitals’ routine tests missed 13 or 22% of these infections. Among seven of those 13 diagnoses made solely by mNGS, results guided targeted treatment with clinical effect.
Redwood City, California-based Karius matches sequences from cfDNA in blood plasma to a curated company database of 21,000 reference microbe genomes and delivers reports that show species occurring in greater than expected concentration. Turnaround is fast—typically about 48 hours from blood draw, including shipping. A team of on-call infectious diseases specialists review results with clinicians, particularly for challenging cases, said Timothy A. Blauwkamp, PhD, Karius’s co-founder and chief scientific officer.
The $2,000 Karius test helps diagnose acute infections in immunocompromised patients, invasive fungal infections, and cardiovascular-related infections, according to the company. A recent study showed that the test identified responsible pathogens in 86% of 15 children with pneumonia and resulted in changes in antibiotics for almost half of them. Meanwhile, standard methods diagnosed only 46% of the children (Diagn Microbiol Infect Dis 2019;94:188–91). Other published data on the first 100 tests used at a children’s hospital showed that sensitivity and specificity of the test for a clinically relevant infection were 92% and 64%, respectively (Open Forum Infect Dis 2019;6:pii:ofz327).
Published evidence is emerging on using the Karius test for diagnosing endocarditis, including a case in which the test helped identify a Coxiella burnetii infection (Open Forum Infect Dis2019;6:ofz242). The company’s website lists several unpublished abstracts it says support use of the test in diagnostic and management algorithms.
Karius is planning to use the test to discern sepsis causes, often missed by culture. A paper describing validation of the test for this purpose compared Karius test results to standard of care on 350 patients with suspected sepsis. The Karius test identified responsible pathogens at a rate about three times higher than blood culture, and 28% higher than all microbiology testing combined. Results from Karius agreed with blood culture 93.7% of the time (Nat Microbiol 2019;4:663–74).
The UCSF and Karius tests may be the most prominent in the mNGS space, but they aren’t the only ones. San Francisco-based IDbyDNA, in partnership with ARUP Laboratories, offers a $500 test of DNA and RNA detecting 900 pathogens in samples from respiratory disease patients. The company delivers results within 29 hours after receiving samples. Robert Schlaberg, MD, PhD, MPH, IDbyDNA’s chief medical officer and co-founder, attributed the quick turnaround to data analysis via the company’s Explify platform, which it markets to other labs.
Tests from Scottsdale, Arizona-based Fry Laboratories use blood and sequence a region of the 16S or 18S rRNA gene found in bacteria, archaea, fungi, protozoa, amoeba, and algae. Tests, which do not use shotgun sequencing, cost $1,495 and generally take a week with shipping, said Jeremy Ellis, PhD, Fry Laboratories’ chief scientific officer. Fry Laboratories compares findings to “a curated NCBI ‘nt’ and ‘16s’ database” about these organisms, added Ellis, who also serves as chief scientific officer at BioID Genomics, a biotechnology company that specializes in sequencing and software solutions to identify and characterize microbes.
mNGS involves many challenges, according to recent reviews of the technology (Clin Infect Dis 2018;66:778–88; Nat Rev Genet 2019;20:341–55). These include sample contamination with nucleic acid during collections and from sequencing reagents, and design of tests for particular pathogens. For example, a test must process RNA to detect RNA viruses. So mNGS pneumonia assays that do not test for RNA might miss respiratory syncytial virus, an RNA virus that commonly causes pneumonia, Chiu said.
Meanwhile, tests do not detect all pathogens equally. For example, mycobacteria might be more difficult to detect because lysing them for nucleic acid release requires more significant cell wall disruption. Also, a negative result might only reflect high leukocyte count of a sample (corresponding to high human DNA and/or RNA host background) or low sequencing depth of a specimen, rather than the absence of a pathogen.
Databases also may contain mislabeled information and include pathogen strains that do not infect humans. Other challenges include differentiating colonization from infection, lack of method standardization, bioinformatic data storage, and patient privacy.
mNGS in Practice
Alexander McAdam, MD, PhD, director of the infectious diseases diagnostic laboratory at Boston Children’s Hospital (BCH) and an associate professor of pathology at Harvard Medical School, said that it’s difficult to fit mNGS into current microbiology testing paradigms built on tests costing less than $100. However, his lab has integrated mNGS into its normal battery of tests. McAdam regularly sends mNGS tests to Karius and UCSF, but only after preliminary negative results on routine tests and approval from a BCH laboratory director. The BCH lab works with ordering clinicians to ensure they understand the tests’ utility and the workflow involved.
At Yale, Peaper occasionally uses mNGS tests for scenarios including diagnostic conundrums, critically ill immunosuppressed patients with tissue-based infections, brain lesions and deep-seated liver abscesses that can’t be biopsied, and suspected endocarditis when valve replacement is not an option. Before ordering, Peaper and clinicians discuss how positive, inconclusive, and negative results would affect care. More data from select patient populations would make decisions easier, he added.
Sometimes results show zero or few reads of the true culprit, so diagnosis requires another method. Peaper recalled recent encephalitis cases detected by antibody, not the UCSF test. Chiu recollected seeing too few reads of Mycobacteriumtuberculosis to call a test positive for it. A different test confirmed M. tuberculosis.
Current mNGS tests can identify diagnostic gaps in hospital lab testing and remind physicians about the breadth of organisms covered in differential diagnoses, said Alex Greninger, MD, PhD. The sensitivity of mNGS is much less affected by antibiotics than that of culture, added Greninger, who participated in early development of the UCSF test and is now developing an mNGS test at University of Washington, where he is an assistant professor of laboratory medicine and associate director of virology.
Greninger pointed out that no mNGS tests have been cleared by the Food and Drug Administration (FDA). Draft FDA guidance issued in 2016 gives labs developing tests an idea of “what validation should look like for a test that’s very different from traditional tests,” he added.
Noting that now-commonplace PCR was new and labor-intensive in the 1980s, Greninger envisions a day when mNGS could become a usual test. “People are showing what’s possible with mNGS,” he said. “We still have to fit it into existing systems in a way that’s realistic about resources available and the value provided.”
Dr. Chiu has a patent on algorithms used in automated software developed by UCSF to analyze and interpret metagenomic sequencing data.
Deborah Levenson is a freelance writer in College Park, Maryland. +Email: firstname.lastname@example.org